Section 1: Getting Started with R

What You Will Need

1. Working and up-to-date installations of R and RStudio
2. Data files: auto.csv and auto.dta (download from course materials)
3. An internet connection
4. Optional: A healthy source of caffeine ☕

Summary

In this section we will dive into R. We start by installing and loading useful packages (dplyr, haven, readr, and pacman). We then load datasets and begin summarizing and manipulating them. Finally, we’ll make our first plots.

---

Packages in R

Open RStudio. Make sure you are running the most recent versions of R and RStudio.

Check Your Versions

• R version: 4.4.2 or newer
• RStudio version: 2024.12.0 or newer

While base R is helpful and powerful, R’s true potential comes from combining the core installation with the many packages generated through open-source collaboration (see CRAN’s list of packages).

Installing Packages

You can check which packages are currently installed using the function installed.packages(). Ready to run your first function? Type installed.packages() into the R console and hit return.

Alternative: Check Packages Tab

You can also check the Packages tab inside RStudio. The checked boxes denote the packages that are currently loaded.

Now, let’s install a few packages that will prove useful this semester. We use the install.packages() function, giving it the name(s) of the package(s) you want to install.

# Install the package named "dplyr"
install.packages("dplyr")

# Install the packages named "haven" and "readr"
install.packages(c("haven", "readr"))

Important Notes:

0. You need an internet connection.
1. The function name install.packages() is plural regardless of the number of packages.
2. Each package’s name is surrounded by quotes (e.g., "haven"). These quotation marks tell R this is a character/string, not an object. Use double quotes for consistency.
3. We can create a vector of packages using c(). Example: c(1, 2, 3) is a three-element vector. Similarly, c("haven", "readr") is a two-element vector of package names. Vectors are a big deal in R.
4. The hashtag (#) creates comments in R.

Loading Packages

To check that installations were successful, we’ll load the packages using the library() function:

library(dplyr)
library(haven)
library(readr)

Note

Notice we didn’t need quotation marks around package names when loading (though it would still work with quotes).

Package Management with pacman

Let’s install another package: the pacman package (yes, like the game! 🎮).

# Install the 'pacman' package
install.packages("pacman")

# Load it
library(pacman)

The pacman package is very meta: it’s a package to manage packages. You can use it to install, load, unload, and update your R packages.

Rather than typing library(...) for every package, use pacman’s p_load() function:

# Old way
library(dplyr)
library(haven)
library(readr)

# New way
p_load(dplyr, haven, readr)

Better yet: If you try to load a package that isn’t installed, p_load() will install it automatically! Let’s install and load ggplot2:

p_load(ggplot2)

Installed and loaded. That’s service! ✨

---

Loading Data

Loading a dataset in R requires three things:

1. The path of the data file (where it exists on your computer)
2. The name of the data file
   
3. The proper function for the file type (e.g., different functions for .csv vs .dta files)

File Paths and Navigation

R wants file paths as character vectors (i.e., the path surrounded by quotations). For example:

"/Users/david/Dropbox/Teaching/MyClass"

Changing and Checking Directories

To change R’s working directory, use setwd():

setwd("/Users/david/Dropbox/Teaching/MyClass")

To find R’s current working directory:

getwd()

Windows Users

When you copy paths from File Explorer, slashes may be backwards (\). Change them to forward slashes (/) or double them (\\).

Navigation Commands

Similar to Linux/Unix:

• .. moves up one level
• ~ returns to your home directory
• You can use full paths or relative paths

# Check current directory
getwd()

# Move up one level
setwd("..")
getwd()

# Go home
setwd("~")
getwd()

# Return to class folder
setwd("Dropbox/Teaching/MyClass")

Smart Path Management

I prefer to store paths at the start of my script for easy updating:

# Path to my class folder
dir_class <- "/Users/david/Documents/MyClass/"

# Path to section 1 folder (inside class folder)
dir_section1 <- paste0(dir_class, "Section01/")

paste0() vs paste()

• paste0() pastes without spaces
• paste() defaults to single space between elements

# Default paste0()
paste0(1, 2, 3)  # "123"

# Default paste()
paste(1, 2, 3)   # "1 2 3"

# Custom separator
paste(1, 2, 3, sep = "+")  # "1+2+3"

RStudio Tab Completion

RStudio assists with completing file paths—begin typing and press Tab. Super useful!

The dir() Function

The dir() function shows contents of a folder:

# Look inside my class folder
dir(dir_class)

# Look inside my section 1 folder
dir(dir_section1)

This helps when you forget file names!

Functions to Load Files

We’ll mostly use:

• readr package: for CSV, TSV, and delimited files
• haven package: for Stata (.dta), SPSS, and SAS files
• Base R: for R’s own formats (.rds, .rdata)

Learn About Functions

To learn more about a function:

1. Press Tab after typing the function name (in RStudio)
2. Type ?function_name in console (e.g., ?read_dta)

Loading a Stata File

Let’s load auto.dta using read_dta() from the haven package:

# Load the .dta file
car_data <- read_dta(paste0(dir_section1, "auto.dta"))

The Assignment Operator <-

The <- operator is central to R. It assigns the value(s) on the right to the name on the left.

When reading aloud, people say “gets”: car_data gets the contents of auto.dta.

To view the data, just type its name:

car_data

Loading a CSV File

For a CSV file, use read_csv() from the readr package:

# Load the .csv file
car_data <- read_csv(paste0(dir_section1, "auto.csv"))

# View it
car_data

Relative Paths

If you’re already in the right directory, you don’t need the full path:

read_csv("Section01/auto.csv")

---

Playing with Data

You now know how to navigate your computer and load data. Let’s do something with those data!

Exploring the Data

Print the data to the console:

car_data

Notice several things:

1. The dataset is a tibble (like a data frame with rules)
2. Dimensions: 74 rows × 12 columns
3. Column types: <chr> (character), <dbl> (double/numeric), <int> (integer)
4. You get a snapshot of the first 10 rows

Getting Column Names

names(car_data)

Viewing First/Last Rows

# First 6 rows
head(car_data)

# First 11 rows
head(car_data, n = 11)

# Last 7 rows
tail(car_data, n = 7)

RStudio Data Viewer

Use View(car_data) or click the dataset in the Environment pane to open RStudio’s data viewer.

Summarizing the Data

Get a quick summary of your dataset:

summary(car_data)

To summarize a single variable, use $ to grab it:

# Grab the price variable
car_data$price

# Summarize price
summary(car_data$price)

Accessing Variables

dataset$variable is how you grab a specific column. RStudio’s tab completion works here too!

Manipulating the Data

The dplyr package offers helpful tools for data manipulation using verbs as actions.

select() - Choose Variables

Keep only variables you care about:

# Select desired variables
car_sub <- select(car_data, price, mpg, weight, length)

# View result
car_sub

We now have 74 rows but only 4 columns!

Alternative: Exclude variables with a minus sign:

select(car_data, -price, -mpg, -weight, -length)

arrange() - Sort Data

Arrange data by one or more columns:

# Arrange by price, then mpg
arrange(car_sub, price, mpg)

Assignment Required!

If you view car_sub now, it’s not arranged. You must assign the result to change the object:

car_sub <- arrange(car_sub, price)

For descending order, use desc():

arrange(car_sub, desc(price), mpg)

summarize() - Create Summaries

Create specific summaries of your data:

# Mean and SD of price
summarize(car_sub, mean(price), sd(price))

# With custom names
summarize(car_sub, price_mean = mean(price), price_sd = sd(price))

For simple summaries, you could also just type:

mean(car_sub$price)
sd(car_sub$price)

Plotting the Data

R’s default plot functions are simple but effective. We’ll cover ggplot2 later, but for now, let’s make quick plots.

Histogram

Create a histogram of miles per gallon:

# Plain histogram
hist(car_sub$mpg)

Make it prettier with labels and a median line:

# The histogram
hist(
  x = car_sub$mpg,
  main = "Distribution of Fuel Economy",
  xlab = "MPG (miles per gallon)"
)

# Add blue line at median
abline(v = median(car_sub$mpg), col = "blue", lwd = 3)

Scatterplot

Plot price vs. mileage:

plot(
  x = car_sub$mpg,
  y = car_sub$price,
  xlab = "Fuel Economy (MPG)",
  ylab = "Price ($)"
)

Clear Code Style

I recommend clearly defining function arguments. It helps keep things straight, as order matters when you’re not naming arguments.

---

Indexing

Nearly everything in R is numerically indexed. You can access individual elements using square brackets [].

Vector Indexing

# Create a vector
x <- c(3, 5, 7, 9)

# Grab the second element
x[2]  # Returns: 5

# Grab second and third elements
x[c(2, 3)]  # Returns: 5 7
x[2:3]      # Same result

# What does 2:3 do?
2:3  # Creates sequence: 2 3

Data Frame Indexing

Data frames have rows and columns: [row, column]

# Grab first row (all columns)
car_sub[1, ]

# Grab first column (all rows)
car_sub[, 1]

# Use column name as index
car_sub[, "price"]

Row, Column Order

Remember: Rows before columns. Leave blank to select all. - [1, ] = first row, all columns - [, 2] = all rows, second column - [3, 4] = third row, fourth column

---

Fun Challenge

Try This

1. What happens if you give head() or tail() a negative n?
2. Can you replicate this behavior using indexing?
3. Can you replicate tail() using only the head() function?

---

Practice Problems

Some classic R-meets-linear-algebra puzzles for your enjoyment:

Problem 1: Identity Matrix

Let I₅ be a 5×5 identity matrix. Demonstrate that I₅ is symmetric and idempotent using R functions.

Hint: Use diag(5) to create an identity matrix.

Problem 2: Idempotent Matrix

Generate a 2×2 idempotent matrix X, where X is not the identity matrix. Demonstrate that X = XX.

Hint: An idempotent matrix satisfies X = X².

Problem 3: Linear Regression

Generate two random variables, x and e, of dimension n = 100 such that x, e ~ N(0, 1).

Generate y according to: y = x + e

Show that regressing y on x using lm() gives estimates: β₀ ≈ 0 and β₁ ≈ 1.

Hint: Use rnorm(100) to generate normal random variables.

Problem 4: Eigenvalues and Trace

Show that if λ₁, λ₂, …, λ₅ are the eigenvalues of a 5×5 matrix A, then:

tr(A) = Σλᵢ (sum of eigenvalues)

Hint: Use eigen() to find eigenvalues and sum(diag()) for the trace.

---

Key Takeaways

Remember

1. Packages extend R’s power - Use pacman::p_load() for easy management
2. File paths are strings - Use paste0() to build paths programmatically
3. Assignment is key - Use <- to save results: new_data <- old_data
4. dplyr verbs - select(), arrange(), summarize() for data manipulation
5. Indexing with [row, column] - Essential for accessing data
6. Use $ for columns - dataset$variable grabs a specific variable

---

Additional Resources

• R for Data Science - Comprehensive guide
• RStudio Cheatsheets - Quick references
• Stack Overflow - Q&A community
• ?function_name - Built-in R help

---

Next: Section 2 - Data Wrangling with dplyr